Context: Organic farming has begun to represent a practical alternative for reducing the environmental impact of crop production and increasing biodiversity. In agricultural systems, the growing demand for stable crop pro-duction requires the adoption of useful tools for the sustainable intensification of agriculture. While organic farming systems face various drawbacks, which affect crop yield. Among the limiting factors, weed management is one of the most difficult aspects of organic farming.Objective: The aim of the study was (i) to evaluate the effect of weed pressure on yield spatial variability (ii) to analyze the spatial variability of yield and yield components based on hand-sampled yield data (iii) to evaluate the ability of the most commonly used vegetation index (NDVI) derived from Sentinel-2 satellite platforms to understand within-field spatial and temporal variability during the crop cycle (iv) to evaluate the ability of the cluster analysis procedures on NDVI satellite data to identify, discriminate and map sub-areas with different yield, yield components, and weed pressure at the farm level in an organic farming system.Methods: Remote sensing techniques (Sentinel-2 images), vegetation index (NDVI), cluster analysis (Hierarchical clustering Ward's minimum variance approach) and hand-sampled data (georeferenced yield and yield com-ponents field data) were used for understanding the spatio-temporal behavior of crops. Results and conclusions: The cluster analysis of NDVI data from Sentinel-2 collected at eight different stages detected the crop spatial and temporal variability at an early stage. In an open field of 24 ha, an area of about 9 ha, showed a high weed level with a yield average value 78 % lower than the most productive cluster area. The yield components, especially the spike number per square meter, also recorded very low values, mainly due to the negative effect of the high presence of weeds (in particular oats), which reached average values of 250 g m-2. An area of 7.3 ha showed the highest yield at harvest with a yield value 45 % higher than the mean harvest data. The study highlights that from the early tillering stage to the booting stage the NDVI maps derived from the Satellite Sentinel-2 and clustered by Ward's method represent the weeds distribution impact on crop yield and yield components and an early warning for the assessment of one of the most important limiting factors in organic farming.Significance: Sentinel-2 data and cluster analysis can improve an efficient assessment and management strategies in an organic farming agricultural system. Furthermore, the use of precision agriculture tools in organic farming can lead to a site-specific management at the farm level and therefore a sustainable intensification of agriculture systems.
Understanding the spatio-temporal behavior of crop yield, yield components and weed pressure using time series Sentinel-2-data in an organic farming system
Marino S.
Primo
2023-01-01
Abstract
Context: Organic farming has begun to represent a practical alternative for reducing the environmental impact of crop production and increasing biodiversity. In agricultural systems, the growing demand for stable crop pro-duction requires the adoption of useful tools for the sustainable intensification of agriculture. While organic farming systems face various drawbacks, which affect crop yield. Among the limiting factors, weed management is one of the most difficult aspects of organic farming.Objective: The aim of the study was (i) to evaluate the effect of weed pressure on yield spatial variability (ii) to analyze the spatial variability of yield and yield components based on hand-sampled yield data (iii) to evaluate the ability of the most commonly used vegetation index (NDVI) derived from Sentinel-2 satellite platforms to understand within-field spatial and temporal variability during the crop cycle (iv) to evaluate the ability of the cluster analysis procedures on NDVI satellite data to identify, discriminate and map sub-areas with different yield, yield components, and weed pressure at the farm level in an organic farming system.Methods: Remote sensing techniques (Sentinel-2 images), vegetation index (NDVI), cluster analysis (Hierarchical clustering Ward's minimum variance approach) and hand-sampled data (georeferenced yield and yield com-ponents field data) were used for understanding the spatio-temporal behavior of crops. Results and conclusions: The cluster analysis of NDVI data from Sentinel-2 collected at eight different stages detected the crop spatial and temporal variability at an early stage. In an open field of 24 ha, an area of about 9 ha, showed a high weed level with a yield average value 78 % lower than the most productive cluster area. The yield components, especially the spike number per square meter, also recorded very low values, mainly due to the negative effect of the high presence of weeds (in particular oats), which reached average values of 250 g m-2. An area of 7.3 ha showed the highest yield at harvest with a yield value 45 % higher than the mean harvest data. The study highlights that from the early tillering stage to the booting stage the NDVI maps derived from the Satellite Sentinel-2 and clustered by Ward's method represent the weeds distribution impact on crop yield and yield components and an early warning for the assessment of one of the most important limiting factors in organic farming.Significance: Sentinel-2 data and cluster analysis can improve an efficient assessment and management strategies in an organic farming agricultural system. Furthermore, the use of precision agriculture tools in organic farming can lead to a site-specific management at the farm level and therefore a sustainable intensification of agriculture systems.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.